Architecture
Neutron AI sits between your application and your model provider as a scoped memory layer. Your platform keeps ownership of users, authorization, provider credentials, billing, and final model calls. Neutron receives only the memory content and scope metadata you choose to send.
flowchart LR
A["Your app or agent"] --> B["Neutron SDK, CLI, API, or MCP"]
B --> C["Scoped memory"]
C --> D["Agent context pack"]
D --> E["Your provider request"]
E --> F["Model provider"]
F --> G["Your product response"]
G --> H["Optional reflection"]
H --> B
Responsibilities
| Layer | Owned by | Responsibility |
|---|---|---|
| Product identity | Your platform | Users, workspaces, roles, billing, and authorization decisions. |
| Neutron scopes | Shared contract | Tenant and scope IDs that keep memory separated by customer, workspace, agent, project, or session. |
| Memory operations | Neutron AI | Remember, recall, forget, compact, and prepare context for agent tasks. |
| Provider requests | Your platform | Model selection, provider credentials, tool policy, final prompts, and response handling. |
| MCP access | Your platform and Neutron AI | You decide which tools a host can call and which scopes the token can access. |
Read Path
sequenceDiagram
participant App as Your app
participant N as Neutron AI
participant M as Model provider
App->>N: Request agent context for a task
N-->>App: Return scoped context pack
App->>M: Send provider request with current user input and context
M-->>App: Return model response
App->>N: Optionally reflect useful outcome
Write Path
Use remember when your product has a durable fact, preference, rule, or summary worth reusing. Use reflect after an agent task when the final outcome reveals something useful for future tasks. Use forget when a user revokes, deletes, or corrects memory.
Integration Boundary
Neutron AI does not need your provider API keys, raw session cookies, payment details, or private application secrets. Keep those inside your own server boundary and pass Neutron only scoped memory content that is necessary for future agent performance.